Fukunaga, Alex


Batch Random Walk for GPU-Based Classical Planning

AAAI Conferences

Graphical processing units (GPUs) have become ubiquitous because they offer the ability to perform cost and energy efficient massively parallel computation. We investigate forward search classical planning on GPUs based on Monte Carlo Random Walk (MRW). We first show experimentally that straightforward parallelizations of MRW perform poorly. Next, we propose Batch MRW (BMRW), a generalization of MRW which performs random walks starting with many seed states, in contrast to traditional MRW which used a single seed state. We evaluate a sequential implementation of BMRW on a single CPU core, and show that a sequential, satisficing planner based on BMRW performs comparably with previous state-of-the-art MRW-based planners. Then, we propose BMRWG, which uses a GPU to perform random walks. We show that BMRWG achieves significant speedup compared to BMRW and achieves competitive performance on a number of IPC benchmark domains.


Classical Planning in Deep Latent Space: Bridging the Subsymbolic-Symbolic Boundary

AAAI Conferences

Current domain-independent, classical planners require symbolic models of the problem domain and instance as input, resulting in a knowledge acquisition bottleneck. Meanwhile, although deep learning has achieved significant success in many fields, the knowledge is encoded in a subsymbolic representation which is incompatible with symbolic systems such as planners. We propose LatPlan, an unsupervised architecture combining deep learning and classical planning. Given only an unlabeled set of image pairs showing a subset of transitions allowed in the environment (training inputs), and a pair of images representing the initial and the goal states (planning inputs), LatPlan finds a plan to the goal state in a symbolic latent space and returns a visualized plan execution. The contribution of this paper is twofold: (1) State Autoencoder, which finds a propositional state representation of the environment using a Variational Autoencoder. It generates a discrete latent vector from the images, based on which a PDDL model can be constructed and then solved by an off-the-shelf planner. (2) Action Autoencoder / Discriminator, a neural architecture which jointly finds the action symbols and the implicit action models (preconditions/effects), and provides a successor function for the implicit graph search. We evaluate LatPlan using image-based versions of 3 planning domains: 8-puzzle, Towers of Hanoi and LightsOut.


Revisiting Immediate Duplicate Detection in External Memory Search

AAAI Conferences

External memory search algorithms store the open and closed lists in secondary memory (e.g., hard disks) to augment limited internal memory. To minimize expensive random access in hard disks, these algorithms typically employ delayed duplicate detection (DDD), at the expense of processing more nodes than algorithms using immediate duplicate detection (IDD). Given the recent ubiquity of solid state drives (SSDs), we revisit the use of IDD in external memory search. We propose segmented compression, an improved IDD method that significantly reduces the number of false positive access into secondary memory. We show that A*-IDD, an external search variant of A* that uses segmented compression-based IDD, significantly improves upon previous open-addressing based IDD. We also show that A*-IDD can outperform DDD-based A* on some domains in domain-independent planning.


Classical Planning in Deep Latent Space: Bridging the Subsymbolic-Symbolic Boundary

arXiv.org Artificial Intelligence

Current domain-independent, classical planners require symbolic models of the problem domain and instance as input, resulting in a knowledge acquisition bottleneck. Meanwhile, although deep learning has achieved significant success in many fields, the knowledge is encoded in a subsymbolic representation which is incompatible with symbolic systems such as planners. We propose LatPlan, an unsupervised architecture combining deep learning and classical planning. Given only an unlabeled set of image pairs showing a subset of transitions allowed in the environment (training inputs), and a pair of images representing the initial and the goal states (planning inputs), LatPlan finds a plan to the goal state in a symbolic latent space and returns a visualized plan execution. The contribution of this paper is twofold: (1) State Autoencoder, which finds a propositional state representation of the environment using a Variational Autoencoder. It generates a discrete latent vector from the images, based on which a PDDL model can be constructed and then solved by an off-the-shelf planner. (2) Action Autoencoder / Discriminator, a neural architecture which jointly finds the action symbols and the implicit action models (preconditions/effects), and provides a successor function for the implicit graph search. We evaluate LatPlan using image-based versions of 3 planning domains: 8-puzzle, Towers of Hanoi and LightsOut.


On Hash-Based Work Distribution Methods for Parallel Best-First Search

Journal of Artificial Intelligence Research

Parallel best-first search algorithms such as Hash Distributed A* (HDA*) distribute work among the processes using a global hash function. We analyze the search and communication overheads of state-of-the-art hash-based parallel best-first search algorithms, and show that although Zobrist hashing, the standard hash function used by HDA*, achieves good load balance for many domains, it incurs significant communication overhead since almost all generated nodes are transferred to a different processor than their parents. We propose Abstract Zobrist hashing, a new work distribution method for parallel search which, instead of computing a hash value based on the raw features of a state, uses a feature projection function to generate a set of abstract features which results in a higher locality, resulting in reduced communications overhead. We show that Abstract Zobrist hashing outperforms previous methods on search domains using hand-coded, domain specific feature projection functions. We then propose GRAZHDA*, a graph-partitioning based approach to automatically generating feature projection functions. GRAZHDA* seeks to approximate the partitioning of the actual search space graph by partitioning the domain transition graph, an abstraction of the state space graph. We show that GRAZHDA* outperforms previous methods on domain-independent planning.


Automatic Extraction of Axioms for Planning

AAAI Conferences

Axioms can be used to model derived predicates in domain-independent planning models. Formulating models which use axioms can sometimes result in problems with much smaller search spaces than the original model. We propose a method for automatically extracting a particular class of axioms from standard STRIPS PDDL models. More specifically, we identify operators whose effects become irrelevant given some other operator, and generate axioms that capture this relationship. We show that this algorithm can be used to successfully extract axioms from standard IPC benchmark instances, and show that the extracted axioms can be used to significantly improve the performance of satisficing planners.


Exploration among and within Plateaus in Greedy Best-First Search

AAAI Conferences

Recent enhancements to greedy best-first search (GBFS) such as DBFS, -GBFS, Type-GBFS improve performance by occasionally introducing exploratory behavior which occasionally expands non-greedy nodes. However, most of these exploratory mechanisms do not address exploration within the space sharing the same heuristic estimate (plateau). In this paper, we show these two modes of exploration, which work across (inter-) and within (intra-) plateau, are complementary, and can be combined to yield superior performance. We then introduces a new fractal-inspired scheme called Invasion-Percolation diversification, which addresses “breadth”-bias instead of the “depth”-bias addressed by the existing diversification methods. We evaluate IP-diversification for both intra- and inter-plateau exploration, and show that it significantly improves performance in several domains. Finally, we show that combining diversification methods results in a planner which is competitive to the state-of-the-art for satisficing planning.


Learning to Prune Dominated Action Sequences in Online Black-Box Planning

AAAI Conferences

Black-box domains where the successor states generated by applying an action are generated by a completely opaque simulator pose a challenge for domain-independent planning. The main computational bottleneck in search-based planning for such domains is the number of calls to the black-box simulation. We propose a method for significantly reducing the number of calls to the simulator by the search algorithm by detecting and pruning sequences of actions which are dominated by others. We apply our pruning method to Iterated Width and breadth-first search in domain-independent black-box planning for Atari 2600 games in the Arcade Learning Environment (ALE), adding our pruning method significantly improves upon the baseline algorithms.


Learning to Avoid Dominated Action Sequences in Planning for Black-Box Domains

AAAI Conferences

Black-box domains where the successor states generated by applying an action are generated by a completely opaque simulator pose a challenge for domain-independent planning. The main computational bottleneck in search-based planning for such domains is the number of calls to the black-box simulation. We propose a method for significantly reducing the number of calls to the simulator by the search algorithm by detecting and pruning sequences of actions which are dominated by others. We apply our pruning method to Iterated Width and breadth-first search in domain-independent black-box planning for Atari 2600 games, adding our pruning method significantly improves upon the baseline algorithms.


Improving Greedy Best-First Search by Removing Unintended Search Bias (Extended Abstract)

AAAI Conferences

Recent enhancements to greedy best-first search (GBFS) improve performance by occasionally adopting a non-greedy node expansion policy, resulting in more exploratory behavior. However, previous exploratory mechanisms do not address exploration within the space sharing the same heuristic estimate (plateau) and the search bias in a breadth direction. In this abstract, we briefly describe two modes of exploration (diversification), which work inter-(across) and intra-(within) plateau, and also introduce IP-diversification, a method combining Minimum Spanning Tree and randomization, which addresses “breadth”-bias instead of the “depth”-bias addressed by the existing methods.